Biology Reference
In-Depth Information
applications. There are three main reasons why the problem of protein stabilization is very
amenable to CPD, and thus became CPD
'
s first application:
1.
Protein stability is correlated with the number of intramolecular interactions occurring
in a protein, and thus can be approximated relatively well by the energy function.
When comparing structures of related mesophilic and thermophilic proteins, the ther-
mophilic variant will often have tighter packing in the hydrophobic core, as well as
more hydrogen bonds and salt bridges on the surface. The simplest strategy to stabilize
a protein is thus to introduce as many additional packing and hydrogen bonding inter-
actions as possible. Fortunately, these types of interactions can be relatively well
approximated by CPD energy functions through their van der Waals and hydrogen
bonding terms.
2.
Native proteins tend to be only marginally stable, and therefore there are usually many
possible mutations that improve stability. A surprising fact of protein biochemistry is
that many natural proteins are only marginally stable, with a
Δ
G of folding between
5
2
and
20 kcal/mole. By comparison, a single hydrogen bond can contribute between
1
2
2
4 kcal/mole to stability,
54
meaning that overall protein stability is often equivalent
to only a few interactions. Explanations for this at-first paradoxical observation have
been offered in detail elsewhere,
55
but this fact suggests that the stability of most pro-
teins can be increased.
3.
Thermostable variants of proteins can be obtained that have virtually unchanged back-
bone structure compared to their mesophilic counterpart. When comparing variants of a
certain enzyme adopted for different temperatures, it is often found that the structures
are virtually identical, despite often having only low sequence identity.
56
This perhaps
surprising finding has favorable implications for the design of thermostable proteins:
after all, if stability (and catalytic activity) at different temperatures can be realized with
the same tertiary structure in many cases, then it should also be possible to create
thermostable versions of other mesophilic proteins without significantly changing the
backbone structure.
and
2
117
These three observations in combination lay out a straight-forward approach to stabilize a
protein through computational design: starting from the structure of the to-be-stabilized
mesophilic protein, use the side-chain placement algorithm to introduce as many favorable
additional intramolecular interactions as possible. The backbone can stay fixed throughout
the calculation. The presumption that the starting protein is only marginally stable implies
that potentially a number of different mutations beneficial for stability can be made, and
CPD energy functions can usually identify these. And indeed, there are many examples in
the literature where this straight-forward approach succeeded in designing variants with
improved stability, often with retained functional properties.
In an early example, Malakauskas et al.
57
were able to increase the stability of a model
protein by 4.3 kcal/mol and shift the T
M
by more than 20
C while maintaining (albeit
reduced) affinity to a binding partner. In this example, seven mutations were introduced
that mostly increased hydrophobic packing. In a comprehensive study by Dantas et al.,
58
nine globular proteins were completely redesigned. On average, only 35% of the wild-type
sequence was retained in the designed variants. Six of these had a T
M
above that of the wild-
type protein. Computational thermostabilization of a functional protein was first described
by Korkegian et al.,
5
who identified a set of three mutations that increased the T
M
of yeast
cytosine deaminase, an enzyme with potential applications in prodrug therapy, by 10
C
while retaining the wild-type catalytic efficiency. Recent work by Borgo et al.
59
introduced
an improved computational protocol that explicitly focuses on packing defects in the
hydrophobic core and identifies mutations that are most likely to fix these defects. Using
this protocol, the authors were able to increase stability of one of the unsuccessful cases
from Dantas et al.
'
s test set by 2.3 kcal/mol.
